• Arabidopsis;
  • genomics;
  • leaf senescence;
  • senescence-associated genes;
  • subtraction suppression hybridization


  1. Top of page
  2. Summary
  3. Introduction
  4. Results and discussion
  5. Experimental procedures
  6. Acknowledgement
  7. References

Leaf senescence is a form of programmed cell death, and is believed to involve preferential expression of a specific set of ‘senescence-associated genes’ (SAGs). To decipher the molecular mechanisms and the predicted complex network of regulatory pathways involved in the senescence program, we have carried out a large-scale gene identification study in a reference plant, Arabidopsis thaliana. Using suppression subtractive hybridization, we isolated approximately 800 cDNA clones representing SAGs expressed in senescing leaves. Differential expression was confirmed by Northern blot analysis for 130 non-redundant genes. Over 70 of the identified genes have not previously been shown to participate in the senescence process. SAG-encoded proteins are likely to participate in macromolecule degradation, detoxification of oxidative metabolites, induction of defense mechanisms, and signaling and regulatory events. Temporal expression profiles of selected genes displayed several distinct patterns, from expression at a very early stage, to the terminal phase of the senescence syndrome. Expression of some of the novel SAGs, in response to age, leaf detachment, darkness, and ethylene and cytokinin treatment was compared. The large repertoire of SAGs identified here provides global insights about regulatory, biochemical and cellular events occurring during leaf senescence.


  1. Top of page
  2. Summary
  3. Introduction
  4. Results and discussion
  5. Experimental procedures
  6. Acknowledgement
  7. References

Leaf senescence, the last stage of development that precedes death, is a genetically programmed process. Senescence program is thus believed to be regulated by specific genes (Buchanan-Wollaston, 1997; Buchanan-Wollaston et al., 2003; Dangl et al., 2000; Gan and Amasino, 1997; Quirino et al., 2000; Smart, 1994; Yoshida, 2003). Expression of many genes associated with photosynthetic activity and other anabolic processes is downregulated during leaf senescence (Gepstein, 1988), whereas others are upregulated. Indeed, several dozen genes upregulated during senescence, designated as senescence-associated genes (SAGs) have been identified in various species (Buchanan-Wollaston and Ainsworth, 1997; Dangl et al., 2000; Lee et al., 2001; Lohman et al., 1994; Nam, 1997). The function of the predicted SAGs' products may give a clue to the biochemical, regulatory, and cellular pathways of senescence. The downregulation of a certain set of genes on the one hand, and the upregulation of others on the other hand, reflect not only the transition from autotrophic to heterotrophic metabolism, but also the induction of de novo synthesis of enzymes participating in the self-destruction process that eventually causes death. Among the prominent SAGs are genes responsible for the execution of the senescence syndrome encoding degradative enzymes: proteinases (Dangl et al., 2000; Lohman et al., 1994; Thompson et al., 1998, 2000), lipases (Thompson et al., 1998, 2000), nucleases (Lers et al., 2001; Rubinstein, 2000), chlorophyllases (Jacob-Wilk et al., 1999), and enzymes for nutrient recycling such as glutamate synthase (Watanabe et al., 1994). Recently, a few senescence-associated regulatory genes have been identified. They are predicted to encode transcription factors (Eulgem et al., 2000; Hinderhofer and Zentgraf, 2001; Robatzek and Somssich, 2002), receptors for senescence perception (Hajouj et al., 2000; Robatzek and Somssich, 2002), and components of intracellular protein trafficking (Guterman et al., 2003). Among the genes that are upregulated during leaf senescence are several whose transcript levels accumulate also under abiotic and biotic stresses (Binyamin et al., 2000; Hanfrey et al., 1996; John et al., 1997; Quirino et al., 1999; Weaver et al., 1998).

While these genes offer insights into the molecular basis of senescence, the number of SAGs identified so far cannot account for the myriad biochemical and cellular events involved in responses to exogenous and endogenous senescence-affecting agents, the operation of multiple signaling cascades, and the execution of the senescence syndrome. Furthermore, the available list of SAGs represents genes identified in a wide range of monocot and dicot species. These species may not share identical regulatory pathways. For example, while leaf senescence of several monocarpic species is controlled by flower and fruit development, such linkage was not found in Arabidopsis thaliana ecotype Landsberg erecta (Nooden and Penney, 2001). Thus, the information accumulated from different unrelated species may not allow the prediction of an integrated network operating during leaf senescence. A global genomic study in a single reference species is therefore required and should provide this information.

Most known SAGs have abundant products and are readily detected and repeatedly reported in the literature. To identify also the non-abundant genes, we used suppression subtractive hybridization (SSH), which was specifically designed for comparing gene expression in different tissues or at different developmental stages (Diatchenko et al., 1996). The SSH method is based on generation of libraries of differentially expressed clones by subtraction of tester cDNA (in our case, senescing leaf) with an excess of driver cDNA (prepared from mature leaf). This PCR-based method includes also a normalization step, which equalizes the abundance of cDNA within the tester population, and a subtraction step that excludes the common sequences. The major advantage of the SSH technology is the enrichment of rarely transcribed clones, and therefore higher sensitivity is attained than by other methods of differential screening.

We present here the most comprehensive gene identification study made, so far, of the monocarpic senescence program in a single reference plant. A large fraction of the identified genes has not yet been studied in the context of senescence, thereby contributing to the unraveling of molecular events regulating this important developmental stage. A more complete inventory of genes in a single reference may also enable the integration of the regulatory pathways of senescence. Furthermore, this information is crucial for identifying targets for the manipulation of leaf senescence.

Results and discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Results and discussion
  5. Experimental procedures
  6. Acknowledgement
  7. References

The SSH approach and its advantages in identifying senescence-related genes

The driver cDNA for the construction of the SSH library was synthesized from mRNA isolated from fully expanded mature green rosette leaves, harvested from plants that had not yet commenced flowering and contained the maximal levels of chlorophyll (Figure 1a,b). The tester cDNA was produced from mRNA isolated from senescent leaves whose chlorophyll levels were around 60% of the initial values. Leaves at this stage contain heterogeneous cell populations, with the oldest cells found at the leaf margins and mature green cells in the central region of the leaf (Figure 1a,b). Thus, a pool of leaves at this stage is a good source for isolating SAGs acting throughout the senescence process.


Figure 1. Effects of age, leaf detachment, darkness, and exogenous application of ethylene and cytokinins on Arabidopsis leaf yellowing.

Development and senescence of whole plants at the indicated week after germination (a) and of individual leaf (b) in the following stages: Y, young; FX, mature fully expanded; ES, early senescence; and during the progressive senescence stages: S1, S2, and S3. Detached leaves (c) were incubated under light or darkness. 1 mm ACC (the ethylene precursor) and 1 µm benzyl adenine (BA) were introduced into the incubation solution of the detached leaves.

Download figure to PowerPoint

A global study of genes whose expression is upregulated during natural senescence of Arabidopsis leaves was carried out by analysis of clones obtained from the SSH libraries. Out of 800 isolated cDNA clones, 350 were analyzed for their sequence and 175 non-redundant sequences were confirmed by Northern blot. Table 1 summarizes Northern blot analyses for 127 cDNA clones. This study revealed about 70 new SAGs induced in Arabidopsis, adding significantly (doubling) to the number of known SAGs. The list also contains around 50 genes that had already been reported, although in various species, as related to senescence. Some of the listed SAGs with predicted functions similar to previously reported SAGs have different Accession numbers. Identification of these known SAGs not only supports previous senescence studies but also confirms the enrichment of senescence-specific clones by SSH. Abundant genes that function in the execution phase of senescence, such as hydrolases, have been repeatedly reported in many species. In our set of SAGs, the proportion of these abundant clones is not higher than that of other genes. These results are not surprising because the SSH method enriches for the non-abundant genes that otherwise would not have been detected. It has been calculated from a large population of enhancer trap lines that about 10% of the total number of genes in Arabidopsis are specifically expressed during senescence so that reported SAGs represent only a limited sample (He et al., 2001). The 127 SAGs (Table 1) represent one-third of the available SSH clones; thus, this library, composed of many yet unidentified genes, is an excellent source for large-scale functional genomics studies.

Table 1.  Differential display of leaf senescence clones
Macromolecular degradation and recycling
 Cysteine protease componentAt1g47128inline image1, 2, 3, 4, 20, 21
 Cysteine proteaseAt4g16190inline image1, 2, 3, 4, 20, 21
 Cathepsin B-like cysteine proteaseAt4g01610inline image1, 2, 3, 4, 20, 21
 Gamma vacuolar processing enzymeAt4g32940inline image7, 19
 AALPAt5g60360inline image1, 2, 3, 4, 20, 21
 APG8At3g15580inline image29, 30
 Ubiquitin carrier proteinAt1g14400inline image6, 28
 26 s Proteasome ATPase subunitAt1g53750inline image28
 SKP1 interacting partner 6At2g21950inline image7
 ATP-dependent Clp protease subunitAt5g53350inline image5
 Endonuclease-putativeAt1g35160inline image 
Amino acid transport
 Amino acid permeaseAt1g58360inline image 
 Cationic amino acid transporter putativeAt3g03720inline image 
 Aspartate amino transferaseAt5g11520inline image10
 Vegetative storage protein 2At5g24770inline image11
 Vegetative storage protein 1At5g24780inline image11
Lipid metabolism
 Lipase-putativeAt2g42690inline image12
 Omega-6 fatty acid desaturaseAt3g12120inline image 
 Phospholipase D putativeAt3g15730NA22, 23
Sugar metabolism
 Beta amylase putativeAt3g23920inline image 
Cell wall
 Glycine rich proteinAt1g30460inline image 
 Beta-1,3 glucanase like proteinAt3g55430inline image18
 Glycine-rich cell wall protein-likeAt4g18280inline image 
 Beta glucosidase-like proteinAt4g27830inline image9, 10
 Nucleotide sugar epimeraseAt4g30440inline image 
 Xyloglucan endo-1,4 beta-d-glucanaseAt4g30270inline image6
 Endo-polygalacturonaseAt4g57510inline image 
 Xylose isomeraseAt5g57655inline image 
 Lipoic acid synthaseAt5g08410inline image 
 Lipoic acid synthaseAt2g20860inline image 
 Acid phosphatase-like proteinAt5g34850inline image 
 Acid phosphatase type 5At3g17790inline image24
 Actin depolymerization factorAt3g46000inline image 
 Metallothionein-like proteinAt1g07600inline image1, 4, 16
 Metallothionein-like proteinAt3g09390inline image1, 4, 16
 Metallothionein 2bAt5g02380inline image1, 4, 16
 Dehydroascorbate reductaseAt1g19570inline image 
 Gluthatione S transferaseAt1g78380inline image4, 25, 26
 Glutathione S transferaseAt5g17220inline image4, 25, 26
 Phospholipid hydroperoxide gluthatione transferaseAt4g11600inline image 
 Cu/Zn superoxide dismutase-like proteinAt5g18100inline image26
 Glyoxalase IIAt1g53580NA8
 Copper Amine OxidaseAt4g12290inline image31
 Ferittin 1 precursorAt5g01600inline image4
 Dioxygenase putativeAt2g25450inline image 
 Catalase 3At1g20620inline image4
Stress, pathogenicity and secondary metabolites
 Myrosinase-binding proteinAt1g52040inline image 
 Aldehyde dehydrogenase precursorAt3g48000inline image8
 Cinnamyl alcohol dehydrogenaseAt1g09500inline image8
 Cinnamyl alcohol dehydrogenase ELI3At4g37990NA8
 Allene oxide cyclaseAt3g25760NA13
 Drought induced cysteine proteinaseAt4g39090inline image 
 Wound-responsive proteinAt1g75380inline image 
 Patatin-like proteinAt2g26560inline image 
 Harpin putativeAt5g53730inline image10
 Nitrilase IIAt3g44300inline image8
 Protein induced upon woundingAt4g24220inline image 
 Limonene cyclase-like proteinAt4g16740inline image 
 Squalene epoxidaseAt5g24160inline image 
 nsLipid transfer protein precursorAt5g59310inline image10
 NADPH oxidoreductase putativeAt1g75280inline image 
 Trypsin inhibitorAt1g73260inline image 
Regulatory genes
 TonB-dependent receptorAt1g05340inline image 
 myo-inositol 1-phosphate synthaseAt2g22240inline image 
 Receptor-like protein kinaseAt5g48380inline image 
 Calcium-binding proteinAt1g18210inline image15
 Cyclic nucleotide-regulated ion channelAt2g23980inline image15, 17
 Calcium-binding proteinAt2g43290inline image15,17
 Calmodulin regulated ion channelAt5g54250inline image15, 17
 Apoptosis inhibitor-putativeAt1g68820inline image 
 RAP2.4-AP2 domain TFAt1g78080inline image 
 Lim domain proteinAt2g39900inline image 
 Auxin responsive GH3-like proteinAt4g37390inline image 
 Auxin-regulated protein putativeAt2g45210inline image 
 G protein beta subunit-like proteinAt2g43770inline image 
 Ras-related small GTP-bindingAt5g47201inline image 
 Zinc finger-like proteinAt3g52800inline image 
 Ring H2 finger proteinAt4g11360inline image 
 Zinc finger-like proteinAt5g10650inline image 
 Small nuclear ribonuclear proteinAt3g07590inline image 
SSH clones involved in various processes
 Dioxygenase putativeAt2g25450inline image 
 No Apical Meristem (NAM)At1g52880inline image6
 Glycine rich RNA-binding proteinAt1g30460inline image 
 Glycine rich RNA binding-putativeAt2g21660inline image 
 Glycine rich proteinAt2g05540inline image 
 Translationally controlled tumor proteinAt3g16640inline image 
 Prephanate dehydrataseAt3g44720inline image 
 NADPH-fettihmprotein reductaseAt4g30210inline image 
 SAG21At4g03280inline image 
LSG clones with unknown functions
 At1g13990inline image 
At1g21670inline image 
At1g30420inline image 
At1g51200inline image 
At1g59870inline image 
At1g63010inline image 
At1g67840inline image 
At1g70900inline image 
At1g71950inline image 
At1g73325inline image 
At1g73750inline image 
At3g02040inline image 
At3g15580inline image 
At3g22600inline image 
At3g25480inline image 
At3g26100inline image 
At3g44100inline image 
At3g51130inline image 
At3g51730inline image 
At3g62550inline image 
At4g11910inline image 
At4g13250inline image 
At4g27020inline image 
At4g35750inline image 
At4g39670inline image 
At5g10860inline image 
At5g19540inline image 
At5g43850inline image 

Most of the cDNA sequences are readily aligned to DNA sequences found in the Arabidopsis database. Less than 50% of the sequences are annotated as unknown or hypothetical. The remaining clones have significant homology to sequences with known or predicted function, allowing us to assign, for each clone, a putative role in the regulation or execution of senescence.

SAGs have been grouped into several categories, based on their predicted function. Examples are: degradation of macromolecules and recycling of metabolites, oxidative metabolism and detoxification of oxygen reactive species, responses to pathogens, biosynthesis of secondary metabolites, and regulation of the initiation and progression of senescence. In the following sections, we outline some of the classes of SAGs.

Macromolecule breakdown: protein degradation and recycling

Senescence-associated genes isolated in our study, which participate in cellular protein degradation processes, represent components of the various known proteolytic systems acting in most subcellular compartments. This group of senescence-associated proteins includes: cysteine proteases, aspartic proteases, and components of the ubiquitin/proteosome system and of the novel autophagic (APG) pathway.

The isolated cDNA clones encoding cysteine proteases found in our study represent different groups. The Arabidopsis aleurain-like protein (AALP) that shares 70% homology with the barley enzyme aleurain and with γ oryzain, both involved in germination, is upregulated also during senescence. Of special interest is our finding of the senescence-related increased expression of a gene encoding a cathepsin B-like cysteine proteinase (At4g01610). Cathepsin B, the animal counterpart, is a major lysosomal cysteine peptidase that has been reported to accumulate during liver aging (Keppler et al., 2000). The observed elevated expression of the vacuolar processing enzyme gene indicates that, in addition to de novo synthesis of proteases during leaf senescence, some of the senescence-related proteases are stored in the vacuole, and are then activated by processing enzymes once senescence is in progress (Kinoshita et al., 1999).

The involvement of ubiquitin-proteasome-pendent proteolysis during leaf senescence is reflected by an increase in the expression of the ubiquitin genes encoding enzymes associated with the ubiquitination cascade (ubiquitin conjugating enzymes). Expression of some of the genes encoding proteasome constituent proteins (At1g53750) also increased during the advance of leaf senescence. This finding suggests that the whole ubiquitin-proteasome pathway is active, at least in Arabidopsis, and contributes to intensive protein degradation (Ingvardsen and Veierskov, 2001). Furthermore, a regulatory role for ubiquitin-dependent proteolysis during senescence has been proposed. A mutation in the gene encoding the ORE9 F-box protein caused a delay in the initiation of leaf rather on the progression of senescence (Woo et al., 2001). The ORE9 F-box protein interacts with a component of the plant SCF complex that controls selective ubiquitination and subsequent proteolysis of targeted proteins. Thus, ORE9 may function in the initiation of senescence and is involved in the degradation, through ubiquitin pathway, of a putative key regulatory repressor of senescence. Another mutant (dLs1), defective in a gene encoding an arginyl-tRNA; proteinarginyl transferase (R-transferase) and involved in the N-end rule proteolytic pathway in yeast and mammals, exhibited delayed senescence. This observation suggests that the N-end rule pathway plays an important role in the progress of leaf senescence (Yoshida et al., 2002).

Our finding that the senescence-related cDNA clone At3g15580 is identical to the APG8 gene suggests its association to a novel autophagic pathway in plants (Doelling et al., 2002). During autophagy, bulk cytosolic constituents and organelles are sequestered in specialized autophagic vesicles and are delivered into the vacuoles for their degradation. This pathway could provide an additional route to protein degradation during senescence. The APG8 gene encodes a component of this autophagic pathway in yeast. An Arabidopsis mutant altered in another component (APG7) of the autophagic pathway displayed premature senescence under nutrient-limiting conditions (Doelling et al., 2002). Recently, a novel autophage gene (AtAPG9) related to senescence has been identified (Hanaoka et al., 2002) but has not yet been found in our subtraction library. Taken together, the results suggest active autophagic recycling during leaf senescence.

Senescence promotes movement of nutrients from the vegetative parts to the fruits or to the seeds. Vegetative storage proteins (VSPs) were suggested to serve as a storage buffer between N losses from senescing leaves and grain filling later in the growth cycle (Rossato et al., 2002). We have observed an increase in the steady-state levels of the VSP1 and VSP2 transcripts during senescence, supporting their postulated function. Before being loaded into the vascular system, amino acids are modified into organic nitrogen compounds, such as the amides glutamine and asparagine. The upregulation of genes encoding glutamine synthase (At1g55090) and aspartate amino transferase (At5g11520) during Arabidopsis leaf senescence is consistent with this notion.

Carbohydrate and lipid metabolism

The upregulation in the levels of β-amylase transcripts (At3g23920) in senescing leaves of Arabidopsis supports previous studies describing the intensive polysaccharide breakdown during senescence. It has been hypothesized that sugars, in addition to their important role in energy supply, might act as signal molecules triggering the senescence program (Yoshida, 2003). The hexokinase-mediated sugar signaling pathway has been suggested to regulate the initiation of senescence (Dai et al., 1999). Lipid degradation also accompanies the other characteristic catabolic processes of leaf senescence (Thompson et al., 1998), and is reflected by an increase in the expression of a lipase (At2g42690) and Phospholipase Dα (Fan et al., 1997). It is most likely that lipid degradation is not just a symptom of the senescence process but may also act in regulating the progression of age-dependent senescence (He and Gan, 2002). They identified an SAG encoding an acyl hydrolase that catalyzes the release of oleic acid from triolein. Overexpression of this gene accelerated senescence whereas antisense suppression inhibited the normal progression of senescence. The release of α-linoleic acid may provide a precursor for the synthesis of jasmonic acid, a senescence-promoting hormone (He et al., 2002). A gene whose product is involved in jasmonic acid biosynthesis and was identified in our SSH library is the allen oxide cyclase (At3g25760) that was also found by He et al. (2002).

Pathogenesis-related proteins and biosynthesis of secondary metabolites

This large group of SAGs consists of genes encoding either signaling/regulatory components or effector proteins that together constitute the plant defense mechanism.

The non-specific lipid transfer protein represented by one of the SSH clones might also be involved in the pathogen defense mechanism. Recent studies indicate that lipid transfer proteins are not involved in intracellular lipid trafficking – the role they were initially proposed to have – but rather in plant resistance to biotic and abiotic stresses (Blein et al., 2002). Their extracellular localization may reflect a need for lipid transfer in the formation of hydrophobic protective layers (cutin and suberin) and in the inhibition of fungal growth. One of the well-characterized defense mechanisms in plants is the hypersensitive response (HR) that is commonly triggered by pathogen attack, causing rapid cell death in the region around the infection site. Our finding of new senescence-related genes that have been shown to participate in the HR confirms previous assumptions (Quirino et al., 1999, 2000) that there is an overlap between pathways leading to cell death in both leaf senescence and HR. The extent and significance of this overlap are, however, not yet clear. The temporal expression pattern of the newly identified HR-related gene, the lethal leaf spot homolog, clearly indicates its exclusive participation in the late stages of the senescence syndrome (Figure 2). The gene encoding myrosinase-binding protein (At1g52040), whose transcript is upregulated during senescence, has been implicated in defense responses (Eriksson et al., 2002). The enzyme myrosinase degrades the secondary compounds glucosinolates upon wounding and their products actively involved in defense against pests. Although, the senescence-associated expression of this gene has not been demonstrated, pathogen-independent induction of several defense genes during senescence has been reported (Quirino et al., 1999) and indicates the involvement of defense genes in the senescence program.


Figure 2. Temporal expression profiles of various SAGs.

RNA gel blots analyses displaying various kinetic patterns.

(a) SAGs with basal expression in pre-senescence stages: (1) Cationic amino acid transporter; (2) amino acid permease; (3) methalothionein.

(b) Early expressed genes: (4) xylose isomerase; (5) RING H2 finger protein.

(c) SAGs displaying transient expression: (6) lipid transfer protein and; (7) calmodulin-like protein.

(d) Late-expressed SAGs: (8) lipoic acid synthase; (9) lls1; and (10) myo-inositol phosphate synthase; cab, chlorophyll a/b-binding protein representing a downregulated photosynthetic gene. 18S rRNA shown in the bottom panel is a loading control. The various senescence stages have been defined in the legends for Figure 1.

Download figure to PowerPoint

Oxidation and detoxification mechanisms

Senescence in general is a regulated oxidative process that involves enhancement in the generation of several reactive oxygen species (ROS; Dangl et al., 2000). Although ROS may have a role in the terminal irreversible phase of senescence, an excess of reactive oxygen might be catastrophic and has to be removed to allow essential biochemical processes, nutrient recycling and transport. The increase in the expression of genes encoding antioxidative defense enzymes prevents possible irreversible damage by ROS in the pre-terminal stages of senescence (Dangl et al., 2000). Cu/Zn superoxide dismutase (At5g17220) is one of the oxygen radical scavenging enzymes that were differentially expressed in senescing stages. The major antioxidants ascorbate and glutathione (Dangl et al., 2000) are produced via the ascorbate–glutathione cycle driven by four enzymes; one of them is the dehydroascorbate reductase whose transcripts are accumulated during senescence (Table 1). Other genes with products involved in detoxification processes are ferritin and metallothionein. The precise role of these two proteins is not clear, but they may have dual roles: detoxification of metal ions released during protein breakdown and/or to function as metal-binding proteins for storage or transport into developing organs (Buchanan-Wollaston and Ainsworth, 1997). Evidence from mammalian cells suggests that metallothioneins protect DNA from damage (Chubatsu and Meneghini, 1993).

Regulatory genes

It is expected that a set of genes may regulate the initiation and/or the rate of progress of the senescence syndrome, and these are major targets for ‘gene hunting’ studies. The initiation and progression of leaf senescence is likely to be regulated primarily at the transcriptional level (He et al., 2001). The SSH method, which has proved to be a powerful tool for the amplification of non-abundant transcripts in other systems (Diatchenko et al., 1999), allowed us to identify a number of new SAGs encoding potential regulators. In addition to the two reported senescence-related receptors, senescence-associated receptor kinase (SARK; Hajouj et al., 2000) and senescence-induced receptor kinase (SIRK; Robatzek and Somssich, 2002), we revealed in our screen a cDNA clone that represents a receptor-like protein kinase (At5g48380). Other mRNAs for signal transduction components preferentially accumulated as leaves senesce are represented by the clones encoding a small GTP-binding protein (At5g47201), related to the oncogene RAS and a G protein beta subunit-like protein (At2g43770).

Another SSH-derived clone encodes an AP2 domain transcription factor, which is identical to RAP2.4 and belongs to the EREBP (ethylene responsive element-binding protein) subfamily. These features make it an attractive candidate for a transcription factor functioning in the leaf senescence program. The gene encoding a putative Lim domain protein (At2g3990). This protein carries a zinc motif called LIM domain and belongs to a family known to participate in transcription and cytoskeleton organization. A study carried out in sunflowers suggested that Lim proteins participate in gene transcription, possibly by assembling and stabilizing transcription complexes (Mundel et al., 2000). A tobacco Lim protein acts as a potential transcription factor in lignin biosynthesis (Kawaoka, 2001). Furthermore, suppression of this Lim protein caused simultaneous reduction in the transcript levels of some phenylpropanoid pathway genes, resulting in low lignin content in transgenic plants.

Other genes whose transcript levels increased during senescence belong to known transcription factor families including a zinc finger-like protein and a RING-H2 finger protein. The present research also demonstrates senescence-induced expression of the genes encoding a calmodulin-regulated ion channel (At5g54250) and calcium-binding protein (At1g18210). Together with our recent observation concerning the involvement of a specific Ca2+-dependent protein kinase in the regulation of leaf senescence (Guterman et al., 2003), these results suggest that Ca2+ may serve as a second messenger in the regulation of leaf senescence. Another interesting SAG represented by clone At1g68820 shows sequence homology to the animal apoptosis inhibitor (IAP) that regulates apoptosis by binding and inhibiting capsases (Lotocki and Keane, 2002). The sequence of the plant gene contains the following: N-glycosylation site, a protein kinase C phosphorylation site, N-myristolation site, and a RING H2 zinc finger. This multidomain sequence and the putative role of the human homolog (IAP) in apoptosis support its possible role as a regulatory gene related to leaf senescence.

An interesting observation is the upregulation of At1g05340, which is also analyzed and annotated by the MIPS A. thaliana database and indicates its association to Ton B-dependent receptor protein (IPR000531). In bacteria, the Ton B protein interacts with outer membrane receptor proteins that carry high-affinity binding and energy-dependent uptake of specific substrates. An example is the active iron transport system involved in iron uptake by bacteria (Braun and Braun, 2002). Although neither the gene nor the protein has been isolated or characterized in plants, they may be implicated in ion transport processes acting mainly during the late developmental stages of leaf senescence.

Temporal patterns of gene expression during leaf senescence

Temporal expression patterns may indicate the role of each gene during the various senescence steps from the initiation signal to the terminal phase of cell death. The kinetics of SAGs may be divided into different general categories (Buchanan-Wollaston, 1997; Dangl et al., 2000), and several of them are considered here (Figures 2 and 3). The first category includes genes that have some degree of basal expression in non-senescent leaves, but their mRNA levels increase markedly during the successive stages of senescence. Among this group are genes encoding cationic amino acid transporters, amino acid permease and metallothionein (Figure 2a). Genes belonging to a second category are expressed during the emergence of the inflorescence stem just prior to or at the onset of senescence and may suggest their participation in the initiation phase of senescence. Among this group are the RING-H2 finger protein and xylose isomerase (Figure 2b). Genes whose products are involved in the terminal phase of leaf senescence during which irreversible loss of cell integrity and viability occurs would display a characteristic profile of late expression (Figure 2d). An example is the lethal leaf spot 1 (lls1) gene, known to be involved in the HR in corn, which is preferentially expressed at the very late phase of the senescence syndrome (Figure 2d, 9). The product of this gene is probably responsible for the irreversible necrosis typically involved in this stage of senescence (Dangl et al., 2000). In contrast to the senescence-enhanced genes, chlorophyll a/b-binding (cab) protein expression displayed the typical temporal pattern of a senescence-downregulated gene reflecting the decline of photosynthesis during senescence (Figure 2).


Figure 3. Temporal expression of senescence-associated regulatory genes.

RNA gel blots analyses displaying various kinetic patterns of the following regulatory genes.

(1) Putative apoptosis inhibitor (At5g68820); (2) putative receptor-like protein kinase (At5g48380); (3) Ras-related small GTP-binding protein (At5g47200); (4) putative Lim domain protein (At2g39900); (5) zinc finger protein (At5g10650). 18S rRNA shown in the bottom panel is a loading control. The various senescence stages have been defined in the legends for Figure 1.

Download figure to PowerPoint

Regulatory genes whose expression is associated with leaf senescence not only are considered to play a crucial role primarily in triggering the onset of senescence but may also have a role in the regulation of various stages throughout the senescence syndrome. Indeed, expression of selected regulatory genes, isolated in the present study, was analyzed and showed different temporal patterns (Figure 3). There are representatives of early genes whose expression is induced at the very early stages of senescence, indicating a regulatory role during the initiation phase. There are genes whose expression is induced at early stages, but thereafter their transcript levels drop. The gene encoding Ras-related small GTP-binding protein displayed such temporal profile (Figure 3). Others (putative apoptosis inhibitor and putative receptor-like protein kinase) are also induced at early stages, but their transcript levels stay high throughout senescence (Figure 3). Genes whose expression is induced very late such as the zinc finger protein may participate in the late senescence stages known to be associated with irreversible necrosis. A biphasic time course was found for the expression of the putative Lim domain protein, suggesting a function both during maturation and later only at the terminal stages of the senescence program (Figure 3).

The expression patterns of senescence-enhanced genes may provide valuable information concerning the sequence of events of the senescence program.

Leaf detachment and hormonal regulation of gene expression during senescence

The fact that leaf senescence is induced and influenced by many different exogenous and endogenous factors implies that there are multiple pathways in the regulation of this process (He et al., 2001). Leaf detachment, for example, causes considerable stress and consequently induces premature senescence. Similar premature senescence is reflected also in the short shelf life of harvested vegetables. Many of the physiological changes occurring during post-harvest senescence, such as chlorophyll loss, deterioration of cellular structures and finally cell necrosis, show similarity to developmental leaf senescence (Figure 1c). To determine the relevance of SAGs identified in attached leaves (Table 1) to post-harvest senescence, the expression pattern of some of the genes was determined in detached leaves and compared with that of attached ones (Figure 4). Three of the tested genes behaved similarly in detached and attached leaves, while NADPH ferrihemoprotein reductase had a different expression profile in the two senescing systems (Figure 4). As there is also a clear distinction in the initiation and progression rate of leaf yellowing under light or in darkness, comparison of gene expression was made with detached leaves incubated either in light or in darkness. Dark treatment that hastened leaf yellowing was as effective in the induction of three of the selected genes NADPH ferrihemoprotein reductase, Gluthatione S-transferase, and a putative protein (clone #636, At4g13250), whereas expression of the lls1 gene was not influenced.


Figure 4. Effects of age, leaf detachment, darkness, and exogenous application of ethylene and cytokinins on the expression of various senescence-associated genes.

Attached leaves were harvested at the indicated stages, homogenized, and analyzed for their gene expression. Detached leaves were subjected to the indicated treatments and gene expression. RNA gel blots for each of the indicated genes are presented. The various senescence stages have been defined in the legends for Figure 1.

Download figure to PowerPoint

The best evidence for hormonal control of leaf senescence is for the hormones ethylene and cytokinins. Many studies in this context have used detached leaves floating or immersed in hormone solutions. Exogenous application of ethylene or its immediate precursor 1-amino-cyclopropane-1-carboxylic acid (ACC; 1 mm) induced premature leaf yellowing and the evolution of endogenous ethylene positively correlated with the progression of this process (Gepstein and Thimann, 1981). It is expected that the effects of ethylene on physiological and biochemical processes of leaf senescence will also be reflected by differential gene expression during this process. Ethylene effect on the expression pattern of several of the senescence-associated genes in detached leaves in the light was studied by exogenous application of the ethylene precursor ACC. The results indicate that 1 mm ACC promotes the expression of all four examined genes (Figure 4). As senescence-enhanced genes depend on developmental signals related to the growth stage as well as many other factors, it may well be that expression of a subset of genes that were not included in this randomly selected sample would be regulated by ethylene. The known delaying effect of cytokinins on physiological characteristics of leaf senescence (Gan and Amasino, 1995) was also reflected by downregulation of the expression of one out of four randomly chosen SAGs (Figure 4). Gluthatione S-transferase showed reduced or delayed expression when leaves were incubated with 1 µm benzyladenine in the dark. The inhibitory effect of cytokinins on the induction of senescence-enhanced genes could not, however, be detected for three other genes that we tested. Taken together, it is evident that hormonal regulation of genes associated with leaf senescence is a complex process. Different defined subsets of genes may be exclusively regulated by only one of the hormones. However, it is not surprising to find that the expression of a common subset of senescence-associated genes is dependent on interaction of two or more different hormones (He et al., 2001). Hormones represent only part of a complex of multiple factors involved in the network of pathways of the senescence program (He et al., 2001). Better understanding of individual pathways and the identification of subgroups of genes that require operation of each specific pathway for expression will allow us to identify common senescence promoter sequences.

Senescence-enhanced gene expression in monocarpic plants as compared to autumn leaves

A recent genomic study aimed at understanding the molecular program of leaf senescence in autumn leaves of the aspen tree (Populus) has yielded large-scale sequencing and analysis of expressed sequence tags (ESTs) obtained by comparison of two cDNA libraries prepared from autumn and young leaves (Bhalerao et al., 2003). The information obtained from this study allows us to compare and estimate the degree of similarity between gene expressions in the two extreme variations of leaf senescence systems; monocarpic senescence as represented by Arabidopsis and autumn leaves of perennial trees. Although no subtractive steps were carried out during the construction of the libraries of aspen, as performed in our present study, the relative EST abundance in the senescing library may provide an approximate indication of transcript levels of individual genes.

The four most abundant ESTs found in aspen autumn leaves are also predominant among the cDNA clones of the Arabidopsis senescing subtraction library. Similar to our finding in Arabidopsis, the most redundant expressed gene is the metallothionein whose transcript was 13 times more abundant in the autumn leaves as compared to young leaves (Bhalerao et al., 2003). In addition to the metallothionein, genes encoding early-light-inducible proteins (ELIP), proteases and components of the ubiquitin degradation pathway are found at the top of the list as the most abundant expressed genes in the aspen autumn leaf. In general, stress-related proteins are preferentially expressed and abundant in Arabidopsis and other monocarpic senescing leaf (Quirino et al., 1999) and are similarly highly expressed in the autumn leaf of aspen (Bhalerao et al., 2003). One example is the ELIP, which is known to accumulate in thylakoids during not only early stages of light-induced greening but also stress (Binyamin et al., 2001). Although ELIP has not as yet been found in our Arabidopsis senescing subtraction library, this gene was shown to be preferentially expressed in senescing leaves of tobacco, a monocarpic plant (Binyamin et al., 2001). Thus, ELIP may prove to play an additional and important role in the senescence program of both perennial trees and monocarpic plants.

Intensive proteolytic degradation is a universal process characteristic of all senescing plant organs and occurs in wide spectrum of species including monocots, dicots, and autumn leaves of perennials (Dangl et al., 2000). The cellular proteolysis in all studied senescing systems is carried out by similar mechanisms including aspartic, Cys proteases and the ubiquitin degradation pathway. The degree of similarity between the senescing systems of monocarpic and autumn leaves is especially high in the family of Cys proteases. Out of the 10 reported genes encoding Cys proteases in aspen (Table 4 in Bhalerao et al., 2003), eight Arabidopsis orthologs have been identified in our study as senescence-regulated genes. Among the common genes are those encoding Cys protease-like protein (At4g16190), Cys protease SAG12 (At5g45890), Protease RD19A (At4g39090), Oryzain (At3g45310), Cathepsin B-like Cys proteinase (At4g01610, At1g02300), Cys Protease RD21A (At1g47128), and AALP (At5g60360). Senescence-enhanced expression of the vacuolar processing enzyme (At4g32940) was also detected in our Arabidopsis study and is not listed as senescence-regulated gene in autumn leaves of aspen (Bhalerao et al., 2003). In general, vacuolar cys proteases are the most predominant proteases in senescing leaves. As the vacuole plays a significant role in defense against pathogens and pests, a defense rather than re-mobilization role has been suggested for several of the senescence-related proteases (Thomas et al., 2003). Vacuolar proteases may not have even a role in the process of chloroplast (and/or other organelle) protein degradation until the very late lytic stages of the vacuole following tonoplast disintegration.

The ubiquitin/proteosome pathway for targeted protein degradation is essential for the control of protein turnover throughout leaf development. The increase in the expression of the ubiquitin-related genes in both monocarpic senescing leaves (Table 1) and autumn leaves of aspen may indicate involvement of ubiquitin in the process of massive protein degradation during senescence. However, in contrast to Arabidopsis, transcript levels of components of the proteosome were not upregulated in autumn aspen leaves (Bhalerao et al., 2003). This result reflects differences between the two senescing systems and may indicate an increase in the proteosome activity exclusively in senescing Arabidopsis leaf and probably in other monocarpic plants, but not in autumn leaves. However, it may also be possible that the accumulated levels of the proteosome components in autumn leaves of aspen are sufficient to breakdown the presumably high levels of ubiquitinated proteins during senescence.

The overall gene expression patterns and the extensive similarity and overlapping in the listed genes of autumn leaves (Bhalerao et al., 2003), of Arabidopsis (Table 1) and other annual senescing leaves (Buchanan-Wollaston et al., 2003) strongly support the notion that the general pattern of leaf senescence metabolism is, to a large extent, similar in a wide range of senescing systems. However, as the senescence syndrome is triggered in perennial autumn leaves by day length shortening, rather than by the development of reproductive organs, it is not surprising that Arabidopsis orthologs of some of the identified aspen senescence-regulated genes are either unknown or not related to senescence (Bhalerao et al., 2003). Similar differences in expression of some of the senescence-enhanced genes are expected even between related monocarpic plants whose senescence is triggered by different signals. For example, the Columbia ecotype (Col-0) of Arabidopsis, which was also studied in the present research, displays a phenomenon characteristic of many monocarpic plants; removal of flower or bolt increases the longevity of the whole plant (Nooden and Penney, 2001). This correlation suggests a link between reproductive organ development and induction of senescence. However, unlike soybean plants whose leaf senescence is controlled by reproductive structures, male and female sterile mutants and surgical removal of inflorescence bolts in Arabidopsis did not increase the longevity of the individual leaves (Nooden and Penney, 2001). Some differences in gene activation related to the induction phase of senescence in various monocarpic species are expected, and comparative genomic studies are required to identify genes related to the induction phase.

In conclusion, we have carried out a comprehensive gene identification study related to a single reference plant representing the monocarpic senescence program. A large number of genes that have not been previously shown to be related to senescence were identified, providing important clues for the molecular understanding of metabolic and regulatory processes associated with the leaf senescence syndrome. The availability of a large inventory of genes, including the non-abundant ones, in a single reference species may also allow integration of the multiple, complex regulatory pathways during senescence. Furthermore, this large repertoire of genes may serve as a rich source for a comprehensive analysis of T-DNA or transposon insertion mutants, each altered in an individual SAG. Tagged mutants are readily available in several Arabidopsis centers. With this reverse-genetic approach, functions for individual SAGs might be identified. Moreover, double, triple or even higher order mutants of different SAGs could be generated, and their phenotypes, related to senescence, may reveal redundant regulatory pathways. The availability of this collection of SAGs would also allow computer analysis of the Arabidopsis genomic sequences in search of common regulatory motifs in the SAG's promoters.

Experimental procedures

  1. Top of page
  2. Summary
  3. Introduction
  4. Results and discussion
  5. Experimental procedures
  6. Acknowledgement
  7. References

Plant material

Following 2–3 days of cold stratification, seeds of A. thaliana Col-0 were germinated and grown on peat pellets (Jiffy 7, Kappa Forenade Well) in a temperature-regulated growth room at 23 ± 1°C with 14-h day/10-h night cycles. For the experiment of the attached leaf senescence, rosette leaves, in position fifth and sixth of each plant, were harvested just before the emergence of the inflorescence stem and were designated as fully expanded mature leaves (FX). Leaves representing various progressive senescence stages were harvested after the inflorescence transition and are presented in Figure 1. For the experiment of detached leaf senescence and hormonal regulation, mature leaves were excised before the emergence of the inflorescence stem and incubated in Petri dishes in darkness or under continuous light at 23 ± 1°C for various periods as indicated.

Leaves were floated either on distilled water or aqueous solutions of 1 mm ACC or benzyl adenine (0.001 mm).

Isolation of RNA and RNA gel blot analysis

Total RNA was prepared according to Hajouj et al. (2000) from rosette leaves. RNA was separated on 1.0% (w/v) agarose formaldehyde gels and blotted to nylon filters (NytranN, Schleicher and Schuell, Dassel, Germany). For RNA blots, universal primers (M13 forward and reverse) were used for PCR amplification and synthesis of the probes. However, if redundancy of sequences has been found in the Arabidopsis databases, specific probes were produced by PCR amplification using specifically designed primers for each of the genes presented in experiments described in Figures 3 and 4. Labeling was performed using the kit rediprime DNA labeling system (RPN1633/1634, Amersham, Uppsala, Sweden). After an overnight hybridization, the membrane was washed, visualized after 2 h with phosphor imager, and exposed to X-ray film.

Suppression subtractive hybridization

Suppression subtractive hybridization was performed with the PCR-Select cDNA Subtraction kit (Clontech Laboratories Inc., Palo Alto, CA, USA) as described by the manufacturer. Two micrograms of senescent leaf mRNA (tester) and 2 µg of fully expanded mature green leaf mRNA (driver) were used. The PCR products generated by the SSH were cloned into the PUC57 vector using the T-cloning kit (MBI-Fermentas, St Leon-Rot, Germany).

Sequencing and homology analysis

Nucleotide sequence of each insert was determined at the Sequencing Services, Technion, Haifa using Dye-deoxy terminators. Sequence homology was analyzed using the blast program.


  1. Top of page
  2. Summary
  3. Introduction
  4. Results and discussion
  5. Experimental procedures
  6. Acknowledgement
  7. References

We thank Dr B. Horwitz for helpful discussions and for critically reading the manuscript.


  1. Top of page
  2. Summary
  3. Introduction
  4. Results and discussion
  5. Experimental procedures
  6. Acknowledgement
  7. References